[Winmerge-svn] SF.net SVN: winmerge: [5507] trunk/Src

[<ki...@us...>]

Revision: 5507
          http://winmerge.svn.sourceforge.net/winmerge/?rev=5507&view=rev
Author:   kimmov
Date:     2008-06-22 10:35:42 -0700 (Sun, 22 Jun 2008)

Log Message:
-----------
TODO: [ 1879979 ] Uni*File::ReadBom() used to check for unicode

Modified Paths:
--------------
    trunk/Src/Common/UniFile.cpp
    trunk/Src/Common/UniFile.h
    trunk/Src/Common/multiformatText.cpp
    trunk/Src/MergeDoc.cpp

Modified: trunk/Src/Common/UniFile.cpp
===================================================================
--- trunk/Src/Common/UniFile.cpp	2008-06-22 12:58:52 UTC (rev 5506)
+++ trunk/Src/Common/UniFile.cpp	2008-06-22 17:35:42 UTC (rev 5507)
@@ -54,6 +54,8 @@
 	m_codepage = getDefaultCodepage();
 	m_txtstats.clear();
 	m_bom = false;
+	m_bUnicodingChecked = false;
+	m_bUnicode = false;
 }
 
 /**
@@ -100,6 +102,20 @@
 	}
 }
 
+/**
+ * @brief Checks if the file is an unicode file.
+ * This function Checks if the file is recognized unicode file. This detection
+ * includes reading possible BOM bytes and trying to detect UTF-8 files
+ * without BOM bytes.
+ * @return true if file is an unicode file, false otherwise.
+ */
+bool UniLocalFile::IsUnicode()
+{
+	if (!m_bUnicodingChecked)
+		m_bUnicode = ReadBom();
+	return m_bUnicode;
+}
+
 /** @brief Record an API call failure */
 void UniLocalFile::LastError(LPCTSTR apiname, int syserrnum)
 {
@@ -304,6 +320,7 @@
 
 	m_bom = bom;
 	m_current = m_data;
+	m_bUnicodingChecked = true;
 	return unicode;
 }
 

Modified: trunk/Src/Common/UniFile.h
===================================================================
--- trunk/Src/Common/UniFile.h	2008-06-22 12:58:52 UTC (rev 5506)
+++ trunk/Src/Common/UniFile.h	2008-06-22 17:35:42 UTC (rev 5507)
@@ -39,6 +39,7 @@
 	virtual String GetFullyQualifiedPath() const = 0;
 	virtual const UniError & GetLastUniError() const = 0;
 
+	virtual bool IsUnicode() = 0;
 	virtual bool ReadBom() = 0;
 	virtual bool HasBom() = 0;
 	virtual void SetBom(bool bom) = 0;
@@ -92,6 +93,8 @@
 	virtual int GetLineNumber() const { return m_lineno; }
 	virtual const txtstats & GetTxtStats() const { return m_txtstats; }
 
+	bool IsUnicode();
+
 protected:
 	virtual bool DoGetFileStatus();
 	virtual void LastError(LPCTSTR apiname, int syserrnum);
@@ -109,6 +112,8 @@
 	int m_codepage; // only valid if m_unicoding==ucr::NONE;
 	txtstats m_txtstats;
 	bool m_bom; /**< Did the file have a BOM when reading? */
+	bool m_bUnicodingChecked; /**< Has unicoding been checked for the file? */
+	bool m_bUnicode; /**< Is the file unicode file? */
 };
 
 /**

Modified: trunk/Src/Common/multiformatText.cpp
===================================================================
--- trunk/Src/Common/multiformatText.cpp	2008-06-22 12:58:52 UTC (rev 5506)
+++ trunk/Src/Common/multiformatText.cpp	2008-06-22 17:35:42 UTC (rev 5507)
@@ -83,7 +83,7 @@
 	UniMemFile ufile;
 	if (ufile.OpenReadOnly(filename))
 	{
-		bIsUnicode = ufile.ReadBom();
+		bIsUnicode = ufile.IsUnicode();
 		ufile.Close();
 	}
 	if (bIsUnicode)
@@ -601,7 +601,7 @@
 BOOL UnicodeFileToOlechar(LPCTSTR filepath, LPCTSTR filepathDst, int & nFileChanged)
 {
 	UniMemFile ufile;
-	if (!ufile.OpenReadOnly(filepath) || !ufile.ReadBom())
+	if (!ufile.OpenReadOnly(filepath) || !ufile.IsUnicode())
 		return TRUE; // not unicode file, nothing to do
 
 	int codeOldBOM = ufile.GetUnicoding();
@@ -699,7 +699,7 @@
 BOOL OlecharToUTF8(LPCTSTR filepath, LPCTSTR filepathDst, int & nFileChanged, BOOL bWriteBOM)
 {
 	UniMemFile ufile;
-	if (!ufile.OpenReadOnly(filepath) || !ufile.ReadBom())
+	if (!ufile.OpenReadOnly(filepath) || !ufile.IsUnicode())
 		return TRUE; // not unicode file, nothing to do
 	int unicoding = ufile.GetUnicoding();
 	// Finished with examing file contents

Modified: trunk/Src/MergeDoc.cpp
===================================================================
--- trunk/Src/MergeDoc.cpp	2008-06-22 12:58:52 UTC (rev 5506)
+++ trunk/Src/MergeDoc.cpp	2008-06-22 17:35:42 UTC (rev 5507)
@@ -1681,9 +1681,10 @@
 	}
 	else
 	{
-		// If file is unicode(including UTF-8 without BOM), try to read BOM 
-		// or else, use the codepage we were given to interpret the 8-bit characters
-		if (encoding.m_unicoding == ucr::NONE || !pufile->ReadBom())
+		// If the file is not unicode file, use the codepage we were given to
+		// interpret the 8-bit characters. If the file is unicode file,
+		// determine its type (IsUnicode() does that).
+		if (encoding.m_unicoding == ucr::NONE || !pufile->IsUnicode())
 			pufile->SetCodepage(encoding.m_codepage);
 		UINT lineno = 0;
 		CString eol, preveol;


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.